This presentation advocates for the strategic integration of symbolic Artificial Intelligence (AI) programming languages, such as Lisp and Prolog, into the public sector's technology stack. The core argument rests on the fundamental distinction between governmental function and commercial operations, emphasizing the constitutional principle of legality (the legales princip). Public administration mandates actions rooted strictly in codified laws, requiring rule-based, symbolic reasoning (analogous to theorem proving) rather than the statistically-driven guessing models prevalent in contemporary AI hype.
The speaker, a Deputy Department Head within the Austrian Federal Chancellery, critiques conventional software development metrics, arguing that true software quality is categorically distinct from measurable quantity and is often found in "massless" ecosystems (e.g., GNU Emacs, Common Lisp). He champions programming paradigms that exhibit "timelessness," promote an architecture of "sparrows on eagles’ backs" (functionality built upon a language engine), and foster a sense of user trust ("in good hands").
The discourse is validated by a confirmed implementation: Austria's "Grants for Companies" e-government service, which uses Common Lisp forms to formalize eligibility criteria from the national transparency database, enabling symbolic reasoning to match companies with suitable grants. Future development targets automatic, declarative explanation generation. The ultimate policy goal is "Law as Code," integrating machine-interpretable symbolic representations directly into the legislative process.
The Need for Symbolic AI Programming Languages in the Public Sector: A Policy and Technical Review
0:13 Public Sector Mandate and Services: Markus Triska, Deputy Department Head in the Austrian Federal Chancellery, oversees key e-government initiatives, including the Business Service Portal (One-Stop Shop for Austrian companies), electronic delivery, and the implementation of the Once-Only Principle (data provision required only once).
1:57 The Advocacy for Symbolic AI: The objective is to highlight the necessity of symbolic AI programming languages (Lisp, Prolog) for public sector applications, even though software development is primarily outsourced.
3:55 The Principle of Legality (Legales Princip): This constitutional principle dictates that all acts of public administration must be rooted in law. Unlike businesses (which can do anything not forbidden), the public sector must do only what is prescribed by law.
5:11 Symbolic vs. Statistical Reasoning: The public sector context necessitates symbolic reasoning (logic, rule application, theorem proving) because laws are known and public; the task is execution and reasoning about existing rules, not guessing laws from data (statistical AI).
8:48 Critique of "Right Tool for the Job": The traditional concept of choosing the "right tool" is viewed as limiting. Programming languages are non-trivial tools; they influence problem perception and are difficult to replace once deployed in an organization.
13:10 Quality over Quantity: Software quality is deemed categorically distinct from quantity (measurement, metrics, telemetry). Examples like GNU Emacs development illustrate high quality achieved through a disregard for typical business metrics (download statistics, cost budgeting, profit).
16:20 Masslessness and Community: High-quality software systems often exhibit "masslessness" (use by a small, expert population). While this creates community benefits (easier organization), it raises valid concerns regarding the availability of skilled programmers (e.g., Lisp programmers) for high-impact government services.
21:24 The "Sparrows on Eagles' Backs" Architecture: High-quality software products (e.g., compilers, Emacs, PostScript interpreters) often utilize a powerful base engine (the "Eagle") and implement substantial functionality (the "Sparrow") on top of that base, often written in the language itself (bootstrapping).
23:56 Homoiconicity Advantage: Homoiconic languages (Lisp, Prolog) are uniquely suited for this architecture in applications due to the ease with which code can be read, reasoned about, and interpreted.
30:00 Timelessness Requirement: Public sector applications require technologies that offer timelessness (stable syntax, long tradition, readily available interpreters/parsers) because e-government services often operate and require updates for 30–40 years.
31:50 Conveying Trust: High-quality software provides the feeling of being "in good hands" (no ads, no tracking, full source code access, high customization). This quality must be reflected in e-government services.
33:39 Suitable Symbolic Languages: Lisp, Prolog, and PostScript are identified as ideal symbolic, homoiconic, and timeless languages suitable for running a public administration.
38:06 Adoption Lag: The slow widespread adoption of a powerful technology (citing the jet engine, which took millennia) is noted as evidence that limited current usage does not negate the intrinsic power of languages like Lisp.
41:00 Confirmed Lisp Use Case (Grants for Companies): The Austrian e-government service "Grants for Companies" (GRS) utilizes Common Lisp to match companies with suitable grants by performing symbolic reasoning on formalized eligibility criteria stored in the Transparency Database.
43:15 Future Development: The service aims to automatically generate human-readable explanations (why a grant is suitable or not) based on the same declarative logic used for the application.
45:28 Security: Using Lisp for the core reasoning component allows for high security, as the engine utilizes a defined language with logical properties that prevent side effects and outside attacks.
50:30 Contradiction Detection: Formalization of laws using symbolic logic allows for the eventual detection of contradictions or inconsistencies within the codified criteria themselves—a potential policy use case for the executive branch.
55:56 "Law as Code": The long-term policy goal is "Law as Code," where machine-interpretable symbolic representations (e.g., Lisp forms) are included directly in legislation, enabling mechanized reasoning and facilitating the generation of human-readable documentation from a single, formal source.